How does shelf information influence consumers ...

1 downloads 0 Views 4MB Size Report
consumers from New South Wales and actual brand market ... Past research (EBI 2007) observing wine shoppers in Australia .... Golden Grove SA 5125.
B u s I n e s s ,

e x p o R t

A n D

m A R k e t I n g

How does shelf information influence consumers’ wine choices? Simone Mueller1, Larry Lockshin1, Jordan Louviere2, Leigh Francis3, Patricia Osidacz3

iNTROdUcTiON

In the last issue of the Wine Industry Journal, we reported on the development of a new method to measure how consumers choose wine, a project funded by the Grape and Wine Research and Development Corporation (Lockshin et al. 2009). We discussed why observing consumers’ choices is more predictive of their real purchases compared with asking consumers how important different attributes are for their purchase decision. A validity test of our online choice experiment resulted in a very high correlation between simulated choices by regular wine consumers from New South Wales and actual brand market shares as reported by AC Nielsen sales data. We also presented the size of the impact we found for different wine labelling and packaging characteristics, such as brand, price, region, alcohol level, closure, label style and label colour, on the choice of a Shiraz wine for a typical dinner with family or friends. While the last article focused on the influence on consumer choice of intrinsic and extrinsic characteristics of the wine itself, this article reports the effect of display information from a simulated retail shelf. While multimedia experiments aimed at forecasting consumer responses to product information have been used for other categories, such as cars or cameras (Urban 1997), we had to develop and test a prototype of shelf information in a simulated wine retail environment. For this first application to wine choice we used a selection of wine display information that might induce consumers to trade up to higher price points. We investigated the response of consumers to shelf display information using sensory descriptions; medals; ratings by wine critics. hOW cONSUMERS ASSESS WiNE

Before presenting details of the study, it is worthwhile reflecting why consumers’ choices might be affected by shelf information displays. Except for highly involved wine drinkers who enjoy the search process, many wine consumers are overwhelmed by the number of wines available to them in a retail store. At the same time, most wine buyers want to make a quick decision. Past research (EBI 2007) observing wine shoppers in Australia found that the average time spent in front of a shelf in a retail outlet was less than a minute, and the total time browsing in the store was about four minutes. Involved shoppers spent up to 15 minutes buying wine, but these were the minority of wine buyers. Many consumers use an implicit logical sequence of 1 Wine Marketing Group, Ehrenberg-Bass Institute for Marketing Science, University of South Australia ([email protected]) 2 Centre for the Study of Choice (CenSoC), University of Technology, Sydney 3 The Australian Wine Research Institute, Adelaide

50

assessing information available to them to simplify the difficult decision to make a choice from the large number of wines that differ in many attributes. These heuristic cues are used to reduce perceived purchase risk by indicating what consumers can expect to get when they purchase a wine. To understand the role of information on consumers’ risk perception it is helpful to think about how wine is actually perceived by consumers. Depending on if and when consumers are able to evaluate a wine characteristic, one can distinguish ‘search’, ‘experience’ and ‘credence’ attributes (Mueller 2004). ‘Search’ characteristics are those that can be assessed before the purchase in front of the shelf. The producer, brand, region, grape variety and packaging are examples of search characteristics. ‘Experience’ characteristics, such as the taste of a wine and whether it is enjoyable, can only be evaluated upon consumption. These are often the main benefits a consumer seeks from purchasing a wine. Other characteristics, such as health effects, environmental benefits, ingredients or production methods used for a wine, which cannot be assessed during consumption, are called ‘credence attributes’.

Many consumers use an implicit logical sequence of assessing information available to them to simplify the difficult decision to make a choice from the large number of wines that differ in many attributes. These heuristic cues are used to reduce perceived purchase risk by indicating what consumers can expect to get when they purchase a wine. When purchasing wine a consumer uses any available ‘search’ information to infer the hidden ‘experience’ and ‘credence’ aspects. While highly involved wine drinkers have a detailed understanding of how wine regions, producers and grape varieties interact and influence how the wine will taste, most consumers have difficulties in understanding these interactions (Lockshin et al. 2006). For instance, they find it very hard to infer wine styles related to regional differences. Less involved wine consumers have been found to remember fewer wine regions or producers overall, with less well-known regions and producers not recalled (Dodd 2005). Hence, those ‘search’ attributes commonly available to a consumer on the shelf are of limited help for most typical wine purchasers in order to reduce their perceived purchase risk.

w w w.w i n e b i z . c o m . a u > M ay/j u n e 2 0 0 9 > V o l 24 n o 3 > w i n e i n d u s t r y j o u r n a l

B u s I n e s s ,

hOW TO REdUcE PERcEiVEd PURchASE RiSK

George A. Akerlof won the Nobel Prize for Economics in 2001 for his breakthrough findings on information asymmetry and its impact on market performance. He found that consumers would pay only a relatively low price when they perceive a purchase to be risky, and that whole markets can fail when the perceived risk is too high, resulting in no transactions. He called these ‘markets for lemons’. We have recently experienced such a market failure in the corporate lending market where to unblock the market, the Government had to reduce banks’ perceived risk by issuing guarantees and by decreasing interest rates. According to Akerlof, this price discount, or ‘risk premium’, can be reduced if market participants provide their trade partners with credible information, which reduces the perceived risk. What information would be able to make a wine purchase decision for consumers less risky if information on the label was of limited help? A sensory description that provides information on the taste the consumer can expect from the wine could be expected to reduce his/her uncertainty and increase the likelihood of a wine being chosen. One only has to make sure that this taste description is understandable, credible and relevant to the consumer, and reflects his or her likely perception of the wine. While consumers can say if they subjectively like a wine, most do not feel very confident to assess its quality from a more objective perspective. So, even when experiencing the wine, its quality, in the sense of degree of excellence, can remain a credence characteristic. While it does not matter when the wine

e x p o R t

A n D

m A R k e t I n g

is drunk privately, there are a large number of social occasions, or when buying wine for a gift, where the buyer wants to make sure that the wine is, indeed, of high quality and is likely to be positively perceived as such by others. For these occasions, a wine consumer could look for expert advice on the objective quality in the form of wine ratings from the store, wine critics’ scores, or medals from wine shows indicating the wine was evaluated favourably by experts compared with other wines. In addition to reducing consumers’ perceived risk, these accolades can also satisfy aspirational needs of exclusiveness or the social acceptance some consumers seek to fulfil with wine (Hall and Lockshin 2000). It is, of course, desirable that these awarded wines are indeed of a higher objective quality for the quality signal to be credible. If these accolades have established social acceptance then they can themselves have a strong influence on how we experience the quality and taste of a wine. Several research studies (Deliza et al. 1996, Deliza and McFie 1996, Guinard et al. 2001, Lange et al. 2002) have demonstrated that consumers can be strongly influenced in their taste evaluation by extrinsic attributes when tasting beverages such as wine or beer in an informed condition. Therefore, wine ratings, show medals and wine critics’ scores can have a combined utility to wine buyers by signalling higher objective quality and also by positively influencing how the wine will taste. A quality signalling mechanism can, of course, lose its reputation over time if it becomes inconsistent or is contrary to expectations. It must be credible and reliable. While we would expect sensory descriptions, wine ratings, show

The Pacifix Muselet • Plain plaque • Printed plaque • Plaqueless • Plain or coloured wire • Fast delivery • Precision component • Australian made Proudly made in Australia by D.J. Young Pty Ltd 710 High Street Kew East VIC 3102 Telephone: +61 3 9859 4468 Fax +61 3 9819 7357 e-mail: sales@pacifi x.com.au

www.pacifix.com.au w i n e i n d u s t r y j o u r n a l > V o l 24 n o 3 > M ay/j u n e 2 0 0 9 > w w w.w i n e b i z . c o m . a u

51

B u s I n e s s ,

e x p o R t

A n D

medals and wine critics’ scores to have a positive influence on consumers’ purchase decisions, little is known about their actual impact and relative importance on wine choice. We, therefore, decided to investigate the effect of these types of information. ShELf diSPLAY iNfORMATiON

As discussed in our last article, we ran two experiments to measure the impact of display information on simulated retail shelves with online choice experiments. One experiment used 21 Shiraz wines, selected from the New South Wales AC Nielsen top 100 sales data, in the price range of $9-$26 with a wide range of sensory properties and which covered both more and less well-known brands. Photographs of these wines were included in a shelf simulation showing five bottles at a time and their real market price. For a subset of six of the 21 wines we also included a short sensory description or a wine critics’ ratings on a simulated shelf talker (see Figure 1). The appearance of both was controlled by a statistical design that allowed us to independently measure the impact of the existence of

m A R k e t I n g

the sensory description and the effect of wine critics’ scores. In total, 365 regular red wine consumers from New South Wales were asked to choose wine for a dinner at home with friends or family using these simulated shelves. The second experiment used graphically simulated wine bottles that varied in six intrinsic wine attributes (brand, country of origin, region of origin, price, price discount and alcohol level) and four extrinsic attributes (label style, label colour, closure and medals). For a more detailed description of this experiment see Lockshin et al. (2009). For the information display we included a quality rating below some of the wines, which had between zero and five stars (see Figure 2). We asked 1233 regular red wine consumers from New South Wales to choose wine from the simulated shelves for a dinner at home with friends and family using these simulated shelves. SENSORY dEScRiPTiONS

Members of the AWRI expert sensory panel characterised the wines, and based on this assessment a short sensory

Figure 1. Example of a screen as part of the study assessing the effect of wine critics’ scores and sensory description in the online choice experiment.

52

A sensory description that provides information on the taste the consumer can expect from the wine could be expected to reduce his/her uncertainty and increase the likelihood of a wine being chosen. One only has to make sure that this taste description is understandable, credible and relevant to the consumer, and reflects his or her likely perception of the wine. While consumers can say if they subjectively like a wine, most do not feel very confident to assess its quality from a more objective perspective. So, even when experiencing the wine, its quality, in the sense of degree of excellence, can remain a credence characteristic. description was formulated for each wine in language understandable to consumers, such as complex Shiraz with leafy and vanilla characters; or dark berries and nuances of chocolate with a smoky aroma. As each description was always combined with the wine it belonged to, we can only measure the effect of the presence or absence of a sensory description, but not which description was more preferred. The impact of the presence of a sensory description was analysed as described previously (Lockshin et al. 2009) by calculating how often a wine was chosen when it had a sensory description compared with when it had none. If a sensory description has a positive influence on choice then wines should be consistently chosen more often with a taste description than with no description. On average over all six wines, the presence of a taste description increased choice by 7.4%. As might be expected, the increase in choice was not the same for all six wines but was found to be always positive and varied between 3.9% and 15.1%.

w w w.w i n e b i z . c o m . a u > M ay/j u n e 2 0 0 9 > V o l 24 n o 3 > w i n e i n d u s t r y j o u r n a l

B u s I n e s s ,

e x p o R t

A n D

m A R k e t I n g

ST4 THE PERFECT CORK FOR PREMIUM

WINES Figure 2. Example of quality star ratings in the online choice experiment with simulated wine bottles.

Further research is necessary to better understand what caused this differential impact on choice. At this stage, we cannot say what the relative contribution of each of the possible aspects of a description is. It might be related to the content and wording of the sensory description; the wine with its unique combination of brand, region and packaging; or the price of the wine. All these variables will have to be combined independently in a new choice experiment to disentangle their individual influences from each other. There are indications that the wording of sensory descriptions used in the marketplace can be improved to be more understandable by consumers. In a recent study, more than a quarter of Australian wine consumers stated that they find it hard to identify flavours indicated on wine back labels when they tasted the wine (Mueller et al. 2009). Non-expert consumers have previously been found to be best able to match wines to short instead of long sensory descriptions (Hughson and Boakes 2002). Non-experts are also better able to match wines to concrete flavour descriptions made by experts than to their own abstract

descriptions (Lawless 1984). But, despite the fact consumers might find simpler taste descriptions easier to understand, it seems to be the case that they find elaborate taste descriptions more appealing when choosing a wine. In the study by Mueller et al. (2009) an elaborate taste description on the back label had, on average, a more positive influence on choice than a simple taste description. More research is required into the optimal translation between the inherent sensory characteristics of a wine and consumer understanding and appeal. WiNE cRiTicS’ ScORES

Wine quality ratings are not widely used in the Australian wine retail market to assist consumer purchase, and there is not such a clear single critic’s influence on the Australian scene compared with markets such as the US, where Robert Parker or the Wine Spectator are very influential. As there are a number of different sources of opinions in Australia, we were not only interested in the effect of lower or higher critics’ scores but also in the effect of the degree of agreement among several critics. On the ‘shelf talker’ of the choice experiment, we displayed three

w i n e i n d u s t r y j o u r n a l > V o l 24 n o 3 > M ay/j u n e 2 0 0 9 > w w w.w i n e b i z . c o m . a u

ACC – Australian Cork Company Pty Ltd 8 Aristotle Close, Industry Park, Golden Grove SA 5125 Ph: +61 8 8251 7609 - Fax: +61 8 8251 7858 E-mail: [email protected]

53

B u s I n e s s ,

e x p o R t

A n D

m A R k e t I n g

Increase in choice (%) Low average rating

High average rating

1.9%

9.8%

(85, 83, 87)

(90, 88, 92)

5.9%

7.2%

(85, 75, 95)

(90, 85, 95)

High agreement

Low agreement

Table 1. Relative impact of wine critic’s point ratings with high and low average and different degrees of agreement between the critics (ratings are shown in brackets).

hypothetical ratings: one indicated to be from Kemenys, one from Vintage Cellars and one from Winestate magazine, with a maximum of 100 points each (see Figure 1). The ratings varied in both their average score (the low average was 85 points and the high one 90 points) and in the degree of agreement between the three scores (low and high agreement), resulting in four conditions in total. Table 1 summarises the average impact of the four rating conditions on relative wine choice. For a low wine rating, where all three sources highly agreed with each other, the impact was, as expected, low (1.9%). Not surprisingly, the condition in which all three rating sources agreed on a high rating had the highest impact, with an average increase in relative choice of 9.8%. One could expect that disagreement between the three scores would signal to the consumers a higher risk. We would then expect ratings with a high variation to have a lower impact on choice than those with lower variance at the same average level. We found that the effect of disagreement differs for the low and high average rating conditions. As expected, the strongly deviating rating scores on the high average had a somewhat lower impact on choice (7.2%) than those agreeing on the same high average (9.8%). Interestingly, if wine raters disagreed on the quality rating of the wines at the lower average level, then consumers seem to be more influenced by the single high score of 95 and hardly consider the very low score. At this stage, we can conclude that high expert wine ratings indeed have a positive impact on consumer choice. For the highest influence on consumer choice, retailers should consider picking the highest score available from different expert ratings and only show several ratings when they agree on a high value (e.g. above 90 points). Rating

Definition

blank

unrated

* ** *** **** *****

an average commercial wine

STAR RATiNGS

With thousands of wines available in Australia, only a relatively small group are rated by external wine experts. However, a retailer could develop its own quality rating system. To investigate the effect of such a retailer-specific system, we integrated a five-star quality rating into the shelf choice experiment with simulated wine bottles (see Figure 3). Before the experiment, respondents were informed about the definition of the quality ratings, from no star to a maximum of five stars for an outstanding wine. Half of all wines in the experiment had no star rating (blank) while 12.5% showed either one, two, four or five stars as a quality rating. As shown in Figure 4, while a wine without any star was chosen 21% of the time, a wine that had a five star rating was chosen 38.6% of the times it appeared. Keeping all other attributes constant, the relative impact on choice from having no rating to a five-star rating was therefore 17.6%. This equates to about a 3.5% increase in relative choice per incremental star. As discussed in more detail in our previous article, choice models allow wine marketers to assess how consumers trade off attributes against each other. An attribute beneficial to consumers, such as a quality rating, could be compensated by an attribute that is less preferred, such as a higher price. While adding a beneficial attribute at a constant price would increase the likelihood of the wine being chosen (i.e. more volume sold) a producer could also consider raising the price by a certain amount. One might also assume that a wine that aims to achieve a five-star rating is more expensive in its production than an average commercial wine. Taking into account the relative choice impact of price, where a decrease of choice by

a commercial wine with above average flavour a well-made fine wine, acceptable for many occasions an excellent wine showing good flavour, structure and balance an outstanding wine, exceeding most others of its type

Figure 3. Definition of star ratings as used in the experiment.

54

w w w.w i n e b i z . c o m . a u > M ay/j u n e 2 0 0 9 > V o l 24 n o 3 > w i n e i n d u s t r y j o u r n a l

B u s I n e s s ,

10.7% was found for an increase from $7.99 to $22.99 as shown in Lockshin et al. (2009), a producer could potentially raise a wine’s price by about $6 if the star rating is increased from four to five stars. Similarly, an additional star from three to four might justify a price increase by about $4.

ShOULd ThE RETAiLER PROVidE cONSUMERS WiTh MORE iNfORMATiON?

From what we found in our experiments, specific information to consumers at the retail outlet has a substantial effect on whether a wine will be selected for purchase. There are some industry voices who suggest that shelf talkers are a lazy form of retailing and that retailers should instead give consumers personal advice. While we clearly advocate a personal retailer consultation that can be tailored to a specific client’s needs, we recommend complementing it with suitable shelf information. While some specialty stores might have a larger number of highly-educated staff, we doubt that the consumer-to-staff ratio and the quick turnover of most staff in liquor outlets would allow an extensive personal conversation with every customer. Nobody would recommend a car seller to not display the engine power, petrol consumption and extra equipment of a car on the information display just because a potential client could also personally ask the salesperson. Moreover, many consumers hesitate to ask for advice as they are afraid to reveal a lack of

A n D

m A R k e t I n g

% of respondents choosing star rating

40%

iNdUSTRY iMPLicATiONS

We found a positive influence on wine choice for all information display types included in the retail shelf simulations. The effects found for the real bottle and the graphical simulated bottle experiments are not exactly comparable, but the overall effect of star ratings was the strongest with a 17.6% increase in choice between no star and five stars, an average of 3.5% per star. Quality ratings in the form of a star seem to be especially suitable as aids to decision-making, presumably as they may be more intuitive and do not require extensive cognitive processing. The presence of sensory descriptions had an average effect of 7.4%, which had a similar impact as found for wine show medals (no medal to Gold and Trophy) with 7.6% (Lockshin et al. 2009, Figure 4) and wine critics’ scores (7-10%). In these experiments, respondents were asked to choose a wine for a dinner with friends or family where there is some degree of social risk present. Results for other purchase occasions are likely to be different. For very special occasions, like a formal dinner or giving wine as a gift, we would expect medals and expert ratings on objective quality to be even more important than measured here. On the other hand, they are likely to be less important for everyday consumption or for drinking wine by oneself. Our results also present a snapshot in time; the effects of medals, stars and scores will be reduced if they are overused or wrongly used and lose their credibility. For quality signals to keep their value they must be used sparingly and consistently to signal high perceived quality. We expect that meaningful and understandable sensory descriptions do reduce purchase risk and, if used wisely, can induce consumers to trade up and try new unknown wines, therefore helping unknown brands to gain market share. More research is necessary to verify this effect.

e x p o R t

38.6%

35%

34.2%

30% 28.1%

Change=17.6%

25% 21.0%

22.1%

20% no Star

1 Star

2 Stars

4 Stars

5 Stars

Star rating

Figure 4. Impact of star ratings on relative choice.

wine knowledge, and our previous research shows that most consumers want to make a quick decision in a minute or two at the most. Appropriate shelf information would help these consumers to make quicker and less risky (to them) choices. We are not advocating putting control in the hands of a few wine critics. Instead, we suggest that wineries provide retailers with as much suitable information as possible to be used for marketing their wines by the retailers. These could be preproduced taste descriptions and information on medals, wine critics’ judgements, and show awards won by the wine. The star ratings were one example that showed that retailers could develop their own in-house rating and wine description system. These currently exist and are used by some retailers. In the long run, the consumer will honour the system that is the most useful to him or her with greater patronage. The wine industry often declares that consumers need to be better educated without specifying the suitable means to do so. There is hardly any more frequented location for consumer information than the retail shelf. Retailers in other product categories, like Amazon, have set examples on how referencing systems - ‘if you like this, then try that’ - can be highly successful in inducing consumers to try new suitable products, trading up and reducing consumers’ perceived risk at the same time. Once we better understand the preference drivers of different consumer segments then wineries and retailers can cross-reference their products for similarity and recommendations. This is just one potential form of consumer information. We as an Australian industry should not miss the chance to meet the strong consumer need for lower perceived risk when buying wines in a crowded and confusing market, because we fear that information from particular sources might create a herd mentality. If we do not wish to provide consumers with what they want, then in the long term we create a window for those importers that are able and willing to do so, or even for other products to satisfy the same consumer needs. fUTURE USE Of iNfORMATiON diSPLAY iN chOicE SiMULATiONS

In this first proof of concept test of integrating information displays into the retail shelf choice method developed by our team, we focused only on a small subset of potential information sources and communication strategies that could be included in choice simulations. The relative impact of different promotional

w i n e i n d u s t r y j o u r n a l > V o l 24 n o 3 > M ay/j u n e 2 0 0 9 > w w w.w i n e b i z . c o m . a u

55

B u s I n e s s ,

e x p o R t

A n D

m A R k e t I n g

materials - neck hangers, environmental messages and even advertising movies (like in YouTube) - can be tested with retail shelf simulations. The method can be adapted to single wine producers, wine regions or the Australian wine industry. It can test consumers in Australia or in overseas markets. In the next phase of our GWRDC-funded project we will apply and refine the method in two US markets. As part of this study, we will test the impact of hypothetical press articles to test the efficiency of different communication devices to influence consumer choice. As in the Australian experiment, we will test the impact of various types of shelf information on US consumers. The results of these experiments will be found on www.winepreferences.com at the end of this year. AcKNOWLEdGEMENTS

This project was supported by Australia’s grapegrowers and winemakers through their investment body the Grape and Wine Research and Development Corporation with matching funding from the Australian Government. We would also like to thank the members of the AWRI expert sensory panel. The members of our industry advisory group and the wineries providing us with photographs are sincerely thanked. REfERENcES Deliza, R. and MacFie H.J.H. (1996) The generation of sensory expectation by external cues and its effect on sensory perception and hedonic ratings: A review. Journal of Sensory Studies 11:103-128. Deliza, R.; MacFie, H.J.H. and Hedderley D. (1996) Information affects consumer assessment of sweet and bitter solutions. Journal of Food Science 61:1080-1084.

56

Dodd, T.H.; Laverie, D.A.; Wilcox, J.f. and Duhan, D.F. (2005) Differential effects of experience, subjective knowledge, and objective knowledge on sources of information used in consumer wine purchasing. Journal of Hospitality and Tourism Research 29:3-19. EBI Ehrenberg-Bass Institute for Marketing Science (2007) Understanding wine shoppers’ purchase decision process and behaviour in store, unpublished research report. Guinard, J.-X.; Uotani B. and Schlich P. (2001) Internal and external mapping of preferences for commercial lager beers: comparison of hedonic ratings by consumers blind versus with knowledge of brand and price. Food Quality and Preference 12:243-255. Hall, J. and Lockshin, L. (2000) Using means-end chains for analysing occasions – not buyers. Australasian Marketing Journal 8(1):45-54. Hughson, A.L. and Boakes, R.A. (2002) The knowing nose: the role of knowledge in wine expertise. Food Quality and Preference 13(7-8), 463472. Lange, C.; Martin, C.; Chabanet, C.; Combris, P. and Issanchou S. (2002) Impact of the information provided to consumers on their willingness to pay for Champagne: comparison with hedonic scores. Food Quality and Preference 13:597-608. Lawless, H.T. (1984) Flavor description of white wine by ‘expert’ and nonexpert wine consumers. Journal of Food Science 40:120-123. Lockshin, L.; Mueller, S.; Louviere, J.; Francis, L. and Osidacz, P. (2009) Development of a new method to measure how consumers choose wine. The Australian and New Zealand Wine Industry Journal 24 (2):35-40. Lockshin, L.; Jarvis, W.; d’Hauteville, F. and Perrouty, J.-P. (2006) Using simulations from discrete choice experiments to measure consumer sensitivity to brand, region, price, and awards in wine choice. Food Quality and Preference 17(3-4):166-178. Mueller, S. (2004) The German wine law from an information economics perspective. International Journal of Wine Marketing 16:72-86. Mueller, S.; Lockshin, L., Saltman, Y. and Blanford, J. (2009) Message on a bottle: the relative importance of wine back label information on choice, working paper University of South Australia. Urban, G.L; Hauser, J.R.; Qualls, W.J.; Weinberg, B.D.; Bohlmann, J.D. and Chicos, R.A. (1997) Information acceleration: validation and lessons from the field. Journal of Marketing Research 34:143-153.

w w w.w i n e b i z . c o m . a u > M ay/j u n e 2 0 0 9 > V o l 24 n o 3 > w i n e i n d u s t r y j o u r n a l